What Made This Part Fail?

Transcription

1 What Made This Part Fail? Richard Palinkas Chemtura Corporation Ride Wheel With Internal Cracks Cracks appeared when the wheel was machined on a lathe. They were not visible on the surface. They were primarily in larger wheels. The material was PPDI cured with 1-4 BD. Mold and resin temperatures were varied with no significant effect. Mix was very good even though this was a hand batch process. Ride Wheel With Internal Cracks Cause And Resolution: The curative was slightly wet, but not enough to create bubbles. Outgassing attempted to produce bubbles as material was gelling. Weak solid allowed pressure to form cracks Very careful drying of the curative and using it shortly after drying resolved the problem. Maintaining strict moisture exclusion prevented further occurrences 1

2 Wheel with Local Bond Loss Wheel had one small area of bond loss, about the size of a palm-print where there was no adhesion to the urethane. There was adhesive remaining on the metal. This happened with about every fifth wheel, but didn t happen at all with the other four wheels. The mold and core temperatures were 260F Wheel With Local Bond Loss What Caused It? The cores were heated in a circulating air oven There were baffles in the sides of the oven to diffuse the air About 5 cores fit in the oven at a time One core was almost always located near a hole in the baffling The input air was hotter than the set temperature of the oven (> 300F) This caused one spot on the core to become overheated and to overheat the adhesive BLOWN OUT RIDE WHEEL Wheel was removed from the ride after 191 days of service. It had blown out. This material was a high performance material. A wheel made with a somewhat lower performance material ran for a whole season with no failure. Material excised from the failed wheel was tested for dynamic properties and was found to have excellent properties. 2

3 Blown Out Ride Wheel Resolution: The wheel design had a concave tread surface with a flat hub. The wheel made from the lower performance material showed evidence of getting very hot internally- yet never went into a melt-down condition. One difference in the dynamic properties of the two materials is that the high performance material maintains modulus to a higher temperature. When the lower performance material got very hot, it lost modulus in the center of the wheel and effectively redistributed the load on either side of center. It became, in effect, a wider wheel. The ride manufacturer redesigned the core with a concave surface to match the tread surface. SOLID WHEEL with GLASS-LIKE CRACKS Wheel is made with a polyether TDI. These cracks developed fairly early into the expected life of the wheel. There is a glass like appearance to the crack surfaces. The cracks grew quickly once they appeared. When the sample was sanded it sanded easily and very smooth. SOLID WHEEL With GLASS-LIKE CRACKS DIAGNOSIS: The stoichiometric ratio was too low. 3

4 Failed Spring Urethane is bonded to core but not metal plate Core is welded to metal plate Part became twisted in service Heating part to 260 F caused part to move back into place Failed Spring Heating part allowed urethane bond to break free and part moved back into place Urethane bond to metal exceeded the yield point of the metal - metal failed Nail Gun Bumper LFPX 2950 / HQEE 98% theory Catalyzed with 33lv cured at 250F Customer designed a in-house test to qualify materials. Test was an accelerated test in modified nail gun. Parts reached temp of 325F. Initial couple of handbatched parts greatly exceeded set specification Machine-made parts were tested and failed randomly sometimes met specifications and sometimes failed Several processing conditions were varied with modest improvement but no effect on consistency 4

5 Nail Gun Bumper Test was not reflective of real life use Inherent variability to test When part was put into field, parts worked Modified test to be more reflective of real use WHEEL with LATERAL CRACKS This is a drive wheel for a baggage handling system. These lateral cracks developed after about one month of service. The original material was a PTMEG-TDI/MBCA. WHEEL With LATERAL CRACKS CAUSE and ENGINEERING SOLUTION: These cracks are typical of drive wheels. There is a stress riser in the contact region due to torque. These are fatigue cracks due to excessive strain cycling. If there is no concern for heat related failure, making the tread thicker should help. Switching to an ester will help with the fatigue. Also, increasing stoichiometry may help. 5

6 Screen with Cracks 90A TDI/MOCA cracked during cure Cracks had rounded edges Only in small aperture screens Cracks would disappear by wiping mold with dry brush before casting Cracks would also disappear by raising mold and material temperature Screen with Cracks Mold release was puddled in the bottom of V shaped sections of mold It rose through liquid urethane Interfacial tensions formed a web between mold sections separating liquid surfaces. This accounts for rounded corners. Pay close attention to amount of mold release and application technique 80A Ester TDI/Moca Part Soft in Middle Case1 Bad part: 61B on Edge 50B in center and drifting Prepolymer temp: 212 F Mold temp: 220 F Catalyzing with Adipic acid Did not see same problem in thinner cross section 6

7 Part Soft in Middle-Case1 Exotherm was high due to elevated prepolymer temperature coupled with catalyst Caused inside to cure softer than outside Lowered prepolymer temperature to 180 F Took out adipic acid Lowered mold temperature to 212 F Switch Tube Made with LF 800 and MOCA Parts are hand batched. Several parts were made same day - some were soft (70-75A) while others were normal 80A Switch Tube 140% stoichiometry Operator miscalculated 7

8 SEAL with PIECES BROKEN OFF The material is TDI PTMEG cured with MBCA. A paper clip can easily pierce it. It dissolved in MEK within minutes. SEAL With PIECES BROKEN OFF CAUSE: The stoichiometric ratio was very low. Part with Swirls Found swirls in finished part on open side When a cross section was taken noted swirls throughout Also noted cracking along swirls 8

9 Part with Swirls NMR analysis indicated ester contamination in ether system Heel of ester remained in tank before being filled with ether Ether and ester incompatible Complete flushing of tank resolved problem Part with Small Crack Small crack noted after machining TDI ether/ MOCA system Part With Small Crack Did not degas liquids When material exothermed bubbles outgassed and split material Resolved problem by degassing material 9

10 Failed Urethane From Wheel Wheel is made with a polyether TDI Part failed early into expected life of wheel Glass-like appearance to polyurethane Noted urethane had powdered on the inside Failed Urethane From Wheel Material was very low in stoichiometry FLAT - SPOTTING TIRE The material is a TDI ester cured with MBCA. Some, but not all, wheels poured with this material had severe flat spotting. Test parts, poured at the same time, had low compression set. The wheels passed dynamometer testing. The customer thought it might be related to a particular lot of material. 10

11 FLAT - SPOTTING TIRE CAUSE and RESOLUTION We found a correlation between the material that flat spotted and drift in hardness- particularly Shore C. Testing the problem lot with samples poured in the lab showed no drift. Severely heat aged material didn t have the problem. Material contaminated with as much as 5% machine flushing solvent showed only a minor problem. A visit to the customer s facility resulted in the discovery that their MOCA was being over heated in air, and was black. Test samples poured with this MOCA had only a minor problem. Test parts poured with the oxidized MOCA and heat aged resin demonstrated the drift phenomenon. The customer installed a new MOCA melter and controlled heat history on prepolymer WHEEL WITH LOST BOND The material was a PPDI Prepolymer. The bond loss occurred during processing. Wheel with Lost Bond The core was heated to 230 F The material was heated to 212 F The mold was 250 F Under these conditions, the exotherm is quite high This material, like many MDIs, goes through a cheesy state The core temperature was too low The cure of the material in contact with the core was slowed compared to the bulk of the material and was in the cheesy state when the bulk of the material had gelled at the peak exotherm and was beginning to shrink. As the bulk of the material contracted, the material next to the core was not strong enough to withstand the force, and broke loose. Raise the core temperature to 275 F and keep the mold temperature at 250 F 11

12 ESCALATOR WHEEL Hardness is 95A. Processor indicated that it was Ether (PTMEG) TDI cured with MBCA. It was used in Korea. It came apart after several months of service. ESCALATOR WHEEL FAILURE ANALYSIS: paper clip wire can be pressed into the surface and pierces it. This is a fairly reliable, simple test for hydrolysis. But the Ether should not have experienced hydrolysis- especially in this short time. Material from the wheel was analyzed for composition. It was an Ester TDI cured with MBCA. It had almost certainly been stored for a long time before use. The processor had not processed an ester in a long time. LESSON: Ask a lot of questions when doing a failure analysis. Question the answers. Test where possible. LAMINATED CROSS-SECTION PART The cross-section of the part had layers like an onion. This is an MDI cured with 1-4 BD. The material was mixed in a meter-mix machine and transferred to the mold by a bucket. The part weighs several pounds. The problem didn t happen all the time. 12

13 LAMINATED CROSS-SECTION PART CAUSE: In transferring the material to the mold sometimes the operators waited much longer than other times. It was observed that the problem occurred when this long wait happened. Some of the material was beginning to gel in the bucket before it was poured into the mold. The partially gelled material didn t knit with the other material in the pour and caused the onion skin. SOLID WHEEL with GLASS-LIKE CRACKS Wheel is made with a polyether TDI. These cracks developed fairly early into the expected life of the wheel. There is a glass-like appearance to the crack surfaces. The cracks grew quickly once they appeared. When the sample was sanded it sanded easily and very smooth. Solid Wheel With Glass - Like Cracks Diagnosis: The stoichiometric ratio was too low (too little curative). 13

14 SEAL WITH PIECES BROKEN OFF The material is TDI PTMEG cured with MBCA. A paper clip can easily pierce it. It dissolved in MEK within minutes. SEAL with PIECES BROKEN OFF CAUSE: At first looked like hydrolysis However, Analysis showed the stoichiometric ratio to be very low. Part Soft in Middle Case 2 80A Ester TDI/MOCA Bad part: 61B on Edge 50B in center and drifting down Prepolymer temp: 212F, Mold temp: 220F Catalyzing with Adipic acid Did not see same problem in thinner cross section 14

15 Part Soft in Middle Case 2 Exotherm was high due to elevated prepolymer temperature coupled with catalyst Caused inside to cure softer than outside Lowered prepolymer temperature to 180F Took out adipic acid Lowered mold temperature to 212F Part Soft in Middle Case 3 MDI ether cured with 1,4 Butanediol Noted softer area in middle when cross section taken Softer area hardened with time exposed to air Changing mold or material temperature did not appreciably solve problem Running at higher stoichiometry resolved problem Part Soft in Middle Case 3 Diagnosis: The Twinkie Effect Occurs most often in MDI ether or ester with thick cross section Moisture from air lower stoichiometry participates in finishing cure Moisture does not penetrate thick cross sections When section is cut that section finishes cure and comes up in hardness - this can happen even after part is several months old Higher stoichiometries help because less free NCO left to react with moisture to finish cure. 15

16 Dual Durometer Part is made with two pieces of LF 750D/MOCA on outside. Adiprene 6060 on inside. Problem was mostly cosmetic: Did not like shrinkage between the top layer of 6060 and the top layer of LF 750D Mold temperature was approximately 200 F Chemical temp within normal range Poured LF 750 onto 6060 during gel Dual Durometer Raised mold temperature to F. Delayed putting LF 750 on until 6060 is through gel point and past liquid state 6060 should be allowed to harden slightly before next layer applied Belt Belt is made with Caytur 21 as curative Part came out soft and cheesy and easily ripped even after postcure 16

17 Belt NMR analysis showed large presence of the plasticizer contained in Caytur Caytur container was not rolled or properly mixed before use, so Caytur separated and a large amount of plasticizer was added with the curative. Bond Failure at Urethane - Urethane Interface Dual durometer 83A/93A Belt wiper MDI/TDI Bond failed between two materials Noted irregular surface at bond line indicating material was soft when second material poured onto it Bond Failure at Urethane - Urethane Interface The first material was soft when second material poured onto it, but not soft enough for materials to mix at the interface When gelling, the MDI material shrinkage pulled it away from the TDI before a good bond could develop Establishing the correct window for pouring the second material is critical 17